Flavodoxin from Anabaena 7120: uniform nitrogen-15 enrichment and

Paul A. O'Farrell, Martin A. Walsh, Andrew A. McCarthy, Timothy M. Higgins, Gerrit ... Brian J. Stockman , Michael D. Reily , William M. Westler , Eld...
3 downloads 0 Views 891KB Size
136

Biochemistry 1988, 27, 136-142

McDonald, C. C., & Phillips, W. D. (1970) in Fine Structure of Proteins and Nucleic Acids (Fasman, G. D., & Timasheff, s. N., Eds.) Dekker, New York. Meadows, D. H., Markley, J. L., Cohen, J. S., & Jardetsky, 0. (1967) Proc. Natl. Acad. Sci. U.S.A. 58, 1307-1313. Neuhaus, D., Wagner, G., VaSZk, M., Kagi, J. H. R., & Wiithrich, K. (1985) Eur. J . Biochem. 151, 257-273. Olejniczak, E. T., Poulsen, F. M., & Dobson, C. M. (1981) J . A m . Chem. SOC.103, 6574-6580. Poulsen, F. M., Hoch, J. C., & Dobson, C. M. (1980) Biochemistry 19, 2597-2607. Redfieid, C., Poulsen, F. M., & Dobson, C. M. (1982) Eur. J . Biochem. 128, 527-531. States, D. J., Haberkorn, R. A., & Ruben, D. J. (1982) J . Magn. Reson. 48, 286-292. Sternlicht, H., & Wilson, D. (1967) Biochemistry 6 , 2881-2892.

Strop, P., Wider, G., & Wuthrich, K. (1983) J . Mol. Biol. 166, 641-667. Sukumaran, D. K., Clore, G. M., Preuss, A., Zarbock, J., & Gronenborn, A. M. (1987) Biochemistry 26, 333-338. Wagner, G., & Wuthrich, K. (1982) J . Mol. Biol. 155, 347-366. Wagner, G., Neuhaus, D., Worgotter, E., VaSfik, M., Kagi, J. H. R., & Wiithrich, K. (1986) Eur. J . Biochem. 157, 275-289. Wedin, R. E., Delepierre, M., Dobson, C. M., & Poulsen, F. M. (1982) Biochemistry 21, 1098-1 103. Wuthrich, K. (1986) N M R of Proteins and Nucleic Acids, Wiley, New York. Wiithrich, K., Billeter, M., & Braun, W. (1984) J. Mol. Biol. 180, 7 15-740. Zuiderweg, E. R. P., Kaptein, R., & Wuthrich, K. (1983) Eur. J . Biochem. 137, 279-292.

Flavodoxin from Anabaena 7 120: Uniform Nitrogen- 15 Enrichment and Hydrogen- 1, Nitrogen- 15, and Phosphorus-3 1 N M R Investigations of the Flavin Mononucleotide Binding Site in the Reduced and Oxidized States? Brian J. Stockman, William M . Westler, Eddie S. Mooberry, and John L. Markley* Department of Biochemistry, 420 Henry Mall, University of Wisconsin-Madison, Madison, Wisconsin 53706 Received July 1, 1981: Revised Manuscript Received September 4, 1981

ABSTRACT: Interactions between flavin mononucleotide ( F M N ) and apoprotein have been investigated in the reduced and oxidized states of the flavodoxin isolated from Anabaena 7120 ( M , -21 000). 'H, 15N, and 31PNMR have been used to characterize the FMN-protein interactions in both redox states. These are compared with those seen in other flavodoxins. Uniformly enriched [ 15N]flav~doxin (>95% isotopic purity) was isolated from Anabaena 7120 grown on K i 5 N 0 3as the sole nitrogen source. 15N insensitive nucleus enhanced by polarization transfer (INEPT) and nuclear Overhauser effect (NOE) studies of this sample provided information regarding protein structure and dynamics. A lH-detected 15Nexperiment allowed the correlation of nitrogen resonances to those of their attached protons. Over 90% of the expected N-H cross peaks could be resolved in this experiment.

Flavodoxins constitute a group of low molecular weight (M, 14 000-23 000), FMNI-containing flavoproteins that mediate electron transfer at low redox potential between the prosthetic groups of other proteins (Mayhew & Ludwig, 1975). In some organisms, flavodoxin is produced constitutively, while in others it is produced only under conditions of limiting iron. Flavodoxins serve as a replacement for the iron-containing protein, ferredoxin, in electron-transfer reactions (Tollin & Edmondson, 1980). The FMN cofactor serves as the redox carrier in flavcdoxins. The coenzyme can exist in three oxidation states, two of which have variable protonation states: oxidized (FMN), one electron Supported by USDA Competitive Research Grant 85-CRCR-l1589. This study made use of the National Magnetic Resonance Facility at Madison, which is supported in part by N I H Grant RR023021 from the Biomedical Research Technology Program, Division of Research Resources. Equipment in the facility was purchased with funds from the University of Wisconsin, the N S F Biological Biomedical Research Technology Program (Grant RR023021), N I H Shared Instrumentation Program (Grant RR02781), and the US. Department of Agriculture. A preliminary account of this research has been presented (Stockman & Markley, 1987). * Author to whom correspondence should be addressed.

0006-2960/88/0427-0136$01.50/0

reduced or semiquinone (FMNH' or FMN'-), and two electron reduced or hydroquinone (FMNH, or FMNH-). The ability for flavin-containing proteins to conduct one- or twoelectron transfers permits them to mediate between oneelectron- and two-electron-transfer pathways. All three redox states of flavodoxin exist in vitro, but redox reactions probably only occur between the reduced and semiquinone states in vivo (Simondson & Tollin, 1980). The redox potentials of both transitions in FMN are altered by its association with the apoprotein. The transition from reduced to semiquinone states in flavodoxin has a midpoint potential of -400 to -500 mV, as compared to -124 m V in free FMN (Simondson & Tollin, 1980; Sykes & Rogers, 1984). Coenzyme in different redox states appears to interact differently with the apoprotein. The redox potential for the 1 Abbreviations: BIRD, bilinear rotation decoupling; COSY, correlated spectroscopy; FMN, flavin mononucleotide; INEPT, insensitive nucleus enhanced by polarization transfer; N M R , nuclear magnetic resonance; NOE, nuclear Overhauser effect; NOESY, nuclear Overhauser spectroscopy; SDS, sodium dodecyl sulfate; Tris, tris(hydroxymethy1)aminomethane; TSP, (trimethy1silyl)propionic acid.

0 1988 American Chemical Society

N M R OF FLAVODOXIN FROM ANABAENA 7 1 2 0

reduced to semiquinone transition could be lowered by destabilizing the reduced form and/or by stabilizing the semiquinone form (Simondson & Tollin, 1980). The ability of the protein to manipulate the redox potential of its cofactor allows nature to fine tune flavodoxins to function in specific reactions. Different types of apoprotein-FMN interactions have been proposed to account for the altered redox potential: ring strain, charge-charge, burial of a charge in a hydrophobic environment, and hydrogen bonding. If a bent conformation of reduced FMN was favored in solution, then holding the reduced coenzyme is a planar orientation would destabilize it (Massey & Hemmerich, 1980). Recent work by Moonen et al. (1984a), however, indicates that reduced FMN exists in a planar conformation in solution. Moonen et al. (1984b) further proposed that interactions between the negative charge on the FMN phosphate and the negative charge on the N1 atom of the reduced isoalloxazine ring could play a major role in altering flavodoxin redox potentials. Hydrogen-bonding changes with redox state have also been proposed to account for the altered redox potentials (Muller, 1972). Multinuclear N M R spectroscopy provides a powerful approach to studies of properties of proteins in solution. N M R studies of larger proteins are facilitated by enriching their spin isotopes that have a low natural abundance: 13Cand lsN. Proteins from cyanobacteria, such as Anabaena 7120, can be enriched economically since the organism can fix nitrogen in the form of nitrate and carbon in the form of carbon dioxide. Thus growth media consisting of K''NO3 and/or 13C02as the sole nitrogen and/or carbon source result in proteins (and all other cell components) uniformly enriched in ISN and/or I3C. As the first step in a multinuclear NMR analysis of Anabaena 7 120 flavodoxin, one- and two-dimensional 'H, 31P,and 15N N M R spectroscopies have been used here to investigate interactions between oxidized and reduced FMN and the protein. This flavodoxin has a molecular weight of 21 000 (SDS-polyacrylamide gel electrophoresis). Recently, Vervoort et al. (1.986) published a comparative I3C, lSN,and 31PNMR study of flavodoxins from four species (Megasphaera elsdenii, Clostridium MP, Azotobacter vinelandii, and Desulfovibrio vulgaris). They discussed their results in terms of hydrogen bonding and the electronic nature of the FMN binding site in oxidized and reduced flavodoxins. Results of the present study are compared and contrasted with their results. Information concerning protein dynamics and side-chain solvent accessibility was obtained from NOE and INEPT ''N NMR experiments. MATERIALS AND METHODS Growth of Cyanobacteria. Seventy-liter batches of Anabaena 7 120 were grown on a slight modification of medium C (Kratz & Myers, 1955). Each batch of cyanobacteria was grown for 6 days under constant illumination and agitation. The temperature was maintained at 23 f 2 OC. Since flavodoxin is produced only under iron-limiting conditions in this organism, the iron content was lowered to 60% (560 kg/L) of that of the original medium. At this iron level, about equal quantities of flavodoxin and ferredoxin were produced. To obtain flavodoxin uniformly enriched with 15N,K1'N03 (98+% isotopic purity) was used as the sole nitrogen source. Flavodoxin Purification. Flavodoxin was purified by the procedure of D. W. Krogmann (personal communication). Typically, 100 mg of flavodoxin was obtained from 600 g of wet cell paste. Protein fractions with A466/A276 ratio greater than 0.14 were considered to be pure. Chemicals. K15N03 (98+%) was purchased from the Mound Facility of the Monsanto Research Corp. Other

VOL.

27, NO. 1, 1988

137

chemicals were reagent grade or better. N M R Spectroscopy. 'H N M R spectra (400 MHz) were obtained in the Fourier transform mode with a Bruker AM400 wide-bore spectrometer. Samples consisted of 1.5 mM flavodoxin in 150 mM phosphate buffer at pH 7.5. The total volume was 0.4 mL in a 5-mm NMR sample tube. Flavodoxin studied in 2 H 2 0was previously lyophilized and resuspended in 2 H 2 03 times. Samples studied in H 2 0 contained 10% 2 H 2 0 to provide the field/frequency lock signal. Reduction of the flavodoxin was accomplished by adding solid sodium dithionite to an argon-flushed solution of flavodoxin in the N M R tube. 'H chemical shifts are referenced to an internal standard of TSP (0.0 ppm). See the figure legends for additional experimental parameters. I5N NMR spectra (50.68 MHz) were obtained on a Bruker AM-500 spectrometer, using a 5-mm broad-band probe. The oxidized ['SN]flavodoxin sample was 4.4 mM in 100 mM phosphate buffer at pH 7.5. The reduced ['SN]flavodoxin sample was 2.1 mM in 100 mM phosphate buffer at pH 7.0. Reduction was achieved as described above. Each sample had a total volume of 0.4 mL in a 5-mm NMR sample tube. The solvent was 90% H20/10% 'H20. "N chemical shifts are referenced to liquid ammonia (0.0 ppm); experimental shifts were determined with respect to an external standard of ('5NH4)2S04in 1 M H N 0 3 , which was assigned a chemical shift of 22.3 ppm. One-dimensional nitrogen observe spectra were recorded of oxidized and reduced [I5N]flavodoxin. Proton-nitrogen couplings were collapsed by using WALTZ-16 (Shaka et al., 1983) decoupling gated on during acquisition only. The acquisition time was 393 ms, followed by a 6-s relaxation delay. The long delay was needed to diminish the NOE on lysine €-amino resonances arising from water saturation that occurs during broad-band decoupling. Nitrogen resonances from both oxidized and reduced [''N]flavodoxin also were observed by using the INEPT pulse sequence. WALTZ- 16 decoupling was used during acquisition to decouple attached protons. The relaxation delay was 6 s. The variable delay was chosen to be 3/8JNH (4.17 ms), with JNH assumed to be 90 Hz. This resulted in singly and triply protonated nitrogens having positive intensities, while doubly protonated nitrogens had negative intensities. The nitrogen spectrum of oxidized ['SN]flavodoxin was obtained with full NOE by power-gated 'H decoupling. Low power (0.5 W) was used to generate an NOE during the 6-s delay time, and WALTZ-16 decoupling was used during acquisition to collapse 'H-ISN coupling. A two-dimensional proton-detected nitrogen experiment was carried out with ['SN]flavodoxin. This experiment, which is known variably as a lsN('H] heteronuclear multiple quantum or lSN('HJ reverse (or inverse) experiment, correlates nitrogen resonances with those of their attached protons and provides higher detection sensitivity of the heteronucleus than direct observation. The pulse sequence used was that described by Muller (1979). The experiment uses a Bruker reverse probe, in which the proton pulses are applied to the inner proton coil, while the heteronuclear pulses are applied to the outer coil. Bruker reverse electronics is also employed. Protons are pulsed from the decoupler, while the standard Bruker transmitter pulses nitrogen. The decoupler and receiver have a common reference frequency in this experiment. A BIRD (Garbow et al., 1982) sequence was employed to eliminate long-range coupling and to suppress signals from protons not coupled to a heteronucleus. "N decoupling was not carried out during acquisition, resulting in an antiphase doublet signal for each

138 B I O C H E M I S T R Y

STOCKMAN ET AL. A

n

A

I

I

0

C N3-H

1

B

I

I

I

11.6

11.0

10.4

Chemical s h i f t h (ppm)

Low-field region of 'H NMR s ectra of flavodoxin in 150 mM phosphate at pH 7.5: (A) reduced in H 2 0 (B) reduced in HzO; (C) oxidized in 2H,0; (D) oxidized in H 2 0 . All spectra are 128 transients, with a 60' pulse and a spectral width of 6024 Hz. The acquisition time was 680 ms, followed by a 2-s relaxation delay. Chemical shifts are referenced to (trimethylsily1)propionic acid. FIGURE 1:

P

correlated resonance. Proton resonances are referenced to TSP, while nitrogen resonances are referenced to liquid ammonia. Referencing was determined by aligning the two-dimensional cross peaks with the corresponding one-dimensional spectra shown to the side and bottom of Figure 5 . 31PN M R spectra (161.98 MHz) were recorded on a Bruker AM-400 wide-bore spectrometer using a IO-" broad-band probe. The sample consisted of 0.8 m M flavodoxin (oxidized or reduced) in 100 m M Tris at pH 8.0 in a total volume of N M R sample tube. Reduction was 2.0 mL in a IO-" achieved as described above. The solvent was 90% H20/ 10% 2H20. 31Pchemical shifts are referenced to 85% phosphoric acid (0.0 ppm).

RESULTS ' H N M R of Reduced and Oxidized Flavodoxin. Figure 1 shows a comparison of the low-field regions of reduced and oxidized flavodoxin in H 2 0 and 2H20. The resonance at 10.5 ppm in oxidized flavodoxin is assigned to the proton on N 3 of the flavin ring via a IH-I5N correlation experiment (see below). Since this resonance was also seen after a 3-day exposure to 2H,0 solution, it must be hydrogen bonded and/or solvent inaccessible in the oxidized state. Changes in the low-field region occur upon reduction, but the resonances have not been assigned. 3 1 PN M R on Reduced and Oxidized Flavodoxin. Figure 2 shows phosphorus N M R spectra of oxidized and reduced flavodoxin. Since only one major resonance is seen in each spectrum, it is clear that it belongs to the phosphate group of the FMN cofactor. The line widths are narrower when protons are decoupled (spectra not shown). The phosphorus chemical shift is indicative of the dianionic form of the monoester (Jones & Katritzky, 1962), but in both redox states the chemical shift is downfield slighlty from that of free FMN. It is clear from the single major resonance observed that this flavodoxin does not have an additional covalently bound phosphate group as has been observed in A . vinelandii flavodoxin (Edmondson & James, 1979). Table I shows a comparison of the phosphorus chemical shifts of five flavodoxins. It can be seen that the flavodoxin from Anabaena 7120 has a greater change in chemical shift

I

I

8.0

4.0

I

Chemical s h i f t

b

I

0.0 (ppm)

-4.0

FIGURE 2: 31PNMR spectra of flavodoxin in 100 mM Tris at pH 8.0 in the (A) reduced and (B) oxidized states. Both spectra are the result of 35 000 transients, obtained with a spectral width of 4854 Hz. The aquisition time was 0.84 s, followed by a 1-s relaxation delay. Chemical shifts are referenced to 85% phosphoric acid.

Table I: Comparison of Phosphorus Chemical Shifts of FMN in Reduced and Oxidized States of Several Flavodoxins 31Pchemical 31Pchemical shift shift (ppm)" (PPm)" flavin reduced oxidized flavin reduced oxidized free FMN Anabaena

5.1 5.9

5.1 5.4

M. elsdenii' D . vulgarisC A . vinelandii?

5.4

5.3

5.5 6.4

5.4

7 120' 6.3 Clostridium 5.8 5.1 MPC "All chemical shifts are relatived to 85% phosphoric acid. bPresent study. 'From Vervoort et al. (1986).

upon reduction than do the other flavodoxins; all five shift in the same direction (downfield) upon reduction. This may be indicative of a small conformational change occurring at the phosphate binding site. It has been demonstrated that distortions in the 0-P-0 bond angle of phosphate esters cause changes in the chemical shift of the phosphorus atom (Gorenstein, 1975). Changes in the electronegativity and *-bonding of the phosphorns also result in chemical shift changes (Letcher & Van Wazer, 1966). The shoulder seen on the phosphorus resonance of the reduced flavodoxin probably arises from a small percentage of flavodoxin in the semiquinone form. This spectrum is similar to that seen for other flavodoxin solutions that have contained small amounts of the semiquinone form (Moonen & Muller, 1982). Since the semiquinone resonance is nearer to the reduced than to the oxidized resonance, the phosphate binding conformation of the semiquinone more closely resembles that of reduced flavodoxin than that of oxidized flavodoxin. ISN N M R on Reduced and Oxidized [lSN]Flavodoxin. One-dimensional nitrogen spectra were obtained of reduced (Figure 3A) and oxidized (Figure 3B) [lSN]flavodoxin. The resonances from flavin ring nitrogens have been assigned in both redox states on the basis of their chemical shifts in free F M N in solution and in other flavodoxins (Moonen et al., 1986). These chemical shifts are given in Table 11. N3 and

VOL. 27, N O . 1, 1988

N M R OF FLAVODOXIN FROM ANABAENA 7120 A

I

A

I

139

B

I

300

I 100

I

200

Chemical s h i f t

h (ppm)

FIGURE 3: I5N NMR direct observation spectra of [15N]fla~cdoxin in the (A) reduced (6000 transients) and (B) oxidized (10500 transients) states. A 90' pulse and a spectral width of 20 833 Hz were used. The acquisition time was 393 ms, followed by a 6-s relaxation delay. Chemical shifts are referenced to liquid ammonia.

N10 in the oxidized protein cannot be assigned in this manner but are distinguishable with an INEPT experiment (see below). Only two resonances are observed in the histidine region (177 and 231 ppm for oxidized and 165 and 229 ppm for reduced flavodoxin). Amino acid analysis (data not shown) suggests the presence of a single histidine residue in this flavodoxin. This indicates that both histidine resonances arise from the same side chain and that no other histidine resonances are expected. The r- and *-resonances cannot be assigned unequivocally on the basis of their chemical shifts in free histidine (Blomberg et al., 1977) because hydrogen bonding to these nitrogens may significantly alter their chemical shifts, as has been observed for the catalytic triad histidine of cy-lytic protease (Bachovchin & Roberts, 1978). Other interesting regions in both spectra are as follows: 105-135 ppm, amide backbone nitrogens and glutamine and asparagine side-chain nitrogens; 84-88 and 72-77 ppm, arginine 6- and q-side-chain nitrogens, respectively; and 3 1-36 ppm, lysine e-amino nitrogens and N-terminal nitrogen. Figure 3 shows that the relaxation delay was sufficient with oxidized flavodoxin to remove a negative N O E on the lysine e-amino nitrogens arising from chemical exhange of saturated water protons. However, the same delay time was not sufficient to remove the negative N O E on the lysine €-amino nitrogens of reduced flavodoxin. Figure 4B shows INEPT spectra of oxidized [15N]flavodoxin. Only resonances from nitrogens with attached protons whose exchange rate with solvent protons is slower than JNH are observed. Thus the lysine €-amino nitrogen, arginine 7side-chain nitrogen, and histidine resonances are not seen. Also, the well-resolved resonance at 147 ppm disappears. This suggests that this signal may be a main-chain proline nitrogen (no attached protons). As can be seen in Figure 4B, resonances of both positive and negative intensity are present in the region from 105 to 120 ppm. This provides a way to distinguish main-chain nitrogens (positive intensity) from asparagine and glutamine side-chain nitrogens (negative intensity) since their

7 160

1 100 Chemical s h i f t

I 40

h (ppm)

I5N NMR spectra of ['5N]flavodoxin: (A) single pulse sequence with applied NOE (868 transients); (B) INEPT pulse sequence (1000 transients). A 90' pulse and a spectral width of 20833 Hz were used in each experiment. The acquisition time for the INEPT experiment was 197 ms, followed by a 6-s relaxation delay. The acquisition time for the NOE experiment was 393 ms, also followed by a 6-s relaxation delay. Chemical shifts are referenced to liquid FIGURE 4:

ammonia. chemical shift regions overlap. Only one of the flavin ring nitrogen resonances of oxidized flavodoxin is seen in the INEPT spectrum. It must correspond to N3, since it is the only protonated nitrogen. This assignment allows the adjacent resonance in the one-dimensional spectrum (Figure 3B) to be assigned to N 10. Figure 4A shows the spectrum of oxidized [15N]flavodoxin with an NOE applied. The lysine side-chain nitrogens show a large and negative NOE. Smaller NOE's are also seen for the N-terminal and asparagine and glutamine side-chain nitrogens. When the spectrum of reduced flavodoxin is compared with this spectrum, it is clear that only a small N O E is being manifested in the reduced spectrum. It is also seen that some, but not all, of the asparagine and glutamine sidechain nitrogens experience an NOE (compare this spectrum with the INEPT spectrum of oxidized flavodoxin in Figure 4B). Figure 5 shows the proton-detected 15Nspectrum for oxidized [15N]flavodoxin. Figure 5 clearly shows that almost all of the resonance overlap, present in the amide proton and nitrogen regions of the respective one-dimensional spectra, has been removed. One also sees cross peaks arising from the arginine b-nitrogens (7.0, 85.0 ppm) and the N 3 nitrogen of the isoalloxazine ring (10.5, 162.5 ppm). This latter cross peak is critical since it allows the assignment of the N 3 proton of oxidized flavodoxin in Figure 1.

DISCUSSION Interactions between the flavin coenzyme and apoflavodoxin are manifested in the chemical shifts of the isoalloxazine ring nitrogens. Hydrogen bonding to these nitrogens changes their resonance position. Pyrrole-type nitrogens, such as N 3 and N10 in oxidized and all nitrogens in reduced FMN, are directly bonded to three other atoms in a plane. This results in the

140 B I O C H E M I S T R Y

STOCKMAN ET AL.

1

ac:

E,

ii I

I

a5 ' H Chemical s h i f t

110

,

60 h (ppm)

FIGURE 5: IH-detected 15Ncorrelation spectrum of oxidized [15N]flavodoxin. Ninety-degree nitrogen and proton pulses were used. The spectral width in each dimension was 4348 Hz. The positive and negative peaks of the antiphase doublets are displayed equivalently here. 'Hchemical shifts are referenced to (trimethylsi1yl)propionic acid; ISNchemical shifts are referenced to liquid ammonia.

lone pair of electrons on the nitrogen being involved in the conjugated n-system of the ring and not being accessible to the surroundings. Pyridine-type nitrogens, however, such as N1 and N5 of oxidized FMN, are bonded to only two atoms. Their lone pairs lie directly in the plane of the ring and do not contribute to the conjugated a-system. Pyridine-type nitrogens thus have exposed lone pairs and are very sensitive to hydrogen bonding; typical chemical shift changes on hydrogen bonding are 5-15 ppm upfield. Pyrrole-type nitrogens are less hydrogen bond sensitive, showing only small downfield shifts of 1-3 ppm (Witanowski et al., 1981). Vervoort et al. (1986) have recently compared FMN nitrogen chemical shifts in various flavodoxins with those for free F M N or FMNH- in polar and for tetraacetylriboflavin (TARF) in apolar environments. The chemical shifts of the isoalloxazine ring nitrogens in these two environments represent the extremes for complete and no hydrogen bonding. Chemical shifts of flavodoxin-bound FMN tend to fall between or near these extremes, allowing the degree of hydrogen bonding to a particular nitrogen to be determined. Table I1 compares the isoalloxazine ring nitrogen resonances of Anabaena 7 120 flavodoxin to those of free F M N and TARF in the oxidized and reduced states. In flavodoxin from Anabaena 7 120, the hydrogen bonding observed is different from any of the other flavodoxins so far studied. N1 appears to form a hydrogen bond in the oxidied state, although it is weaker than those in the other types of flavodoxin. In the reduced state, the chemical shift is dominated by the negative charge present, as in the other flavodoxins. N 3 in the oxidized flavodoxin appears to form a stronger hydrogen bond than those of the other flavodoxins, as evidenced by its downfield chemical shift. It also appears to form a stronger hydrogen bond in the reduced state than in the other flavodoxins. The existence of this hydrogen bond in the oxidized state is also shown by the lack of exchange in 2Hz0 of the N3 proton (see Figure 1). N5 shows the greatest difference from the other flavodoxins. In the oxidized protein, the chemical shift is at least 6 ppm farther upfield than that of any of the other flavodoxins. Its chemical shift is almost

Table 11: Comparison of I5N Chemical Shifts of FMN in Anabaena 7 120 Flavodoxin with Those of Model Compounds" Oxidized Flavin atom Anabaena 7120' FMNC TARF~ N1 188.0 190.8 199.9 N3 162.5 160.5 159.8 N5 335.0 334.7 344.3 N10 163.5 164.6 150.2 Reduced Flavin atom Anabaena 7 120' FMNH-C TARFH,~ N1 182.5 182.6 116.7 N3 151.5 149.3 145.8 N5 55.0 57.7 60.4 N10 95.6 91.2 72.2 "All I5N chemical shifts are relative to liquid ammonia and are expressed in ppm. 'Present study. The solvent was 100 mM potassium phosphate; the pH was 7.5 for oxidized and 7.0 for reduced flavodoxin. CFrom Vervoort et al. (1986). The solvent was 100 mM potassium pyrophosphate at pH 8.0. dFrom Vervoort et al. (1986). TARF and TARFH2 are the oxidized and reduced forms of tetraacetylriboflavin, respectively. The solvent was CHC1,.

the same as that for free FMN in solution. In the reduced flavodoxin, N5 is also shifted upfield by more than 6 ppm as compared to the other flavodoxins and again resonates near free FMNH- in solution. The fact that N3 remains strongly hydrogen bonded in both redox states could be explained simply by having a functional group of the protein nearby with which to hydrogen bond. Since N 3 remains protonated in each redox state, the same functional group could be hydrogen bonding in both cases. This cannot be true for N5, however. The protonation state of N 5 varies with redox state. N5 is a hydrogen bond donor in the reduced state and a hydrogen bond acceptor in the oxidized state. Three explanations can be considered: (1) Hydrogen bonding in both redox states is to a group on the protein. If so, the group must be different in each redox state. The hydrogen-bonding change could be effected by changes in the electrostatic environment induced by a redox-related conformational change that results in different functional groups being located in the vicinity of N5. The X-ray crystal structure data for the smaller molecular weight Clostridium MP flavodoxin (Ludwig et al., 1975) indicate that only small conformational changes occur upon reduction. (2) Hydrogen bonding in one redox state is to protein and in the other redox state is to water. Since the dimethylbenzene edge of FMN is known from crystal structures to be exposed to solvent in Clostridium MP (Burnett et al., 1974), D . vulgaris (Watenpaugh et al., 1972), and Anacystis nidulans flavodoxin (Smith et al., 1983), this may be reasonable. Also, in Clostridium M P flavodoxin, the proton on N 5 of the semiquinone forms a hydrogen bond to the protein (Smith et al., 1977). Thus in Anabaena 7120 flavodoxin, N 5 may be hydrogen bonded to the protein in the reduced state and to water in the oxidized state. (3) Hydrogen bonding in both redox states is to water. N5 may be solvent accessible in both redox states. The INEPT and NOE-applied spectra of oxidized flavodoxin give a qualitative description of the lysine, glutamine, and asparagine side-chain environments. The siie of an NOE effect depends on, among other things, the correlation time of the nitrogen-proton internuclear vector. Since the correlation times for side-chain nitrogens exposed to solvent can be considerably shorter because of their increased mobility and/or increased proton exchange rates relative to those nitrogens buried within the protein, the observed negative NOE's probably indicate that the lysine side-chain nitrogens and some of the asparagine and glutamine side-chain nitrogens are on

VOL. 2 7 , N O . 1 , 1988

N M R O F FLAVODOXIN FROM ANABAENA 7120

the surface of the protein. This is also true for the N-terminal nitrogen, In the INEPT experiment, no signals are observed for the lysine side-chain nitrogens because the protons are in fast exchange (lifetime